/go-error-handling

name: go-error-handling
description: Use when writing Go code that returns, wraps, or handles errors — choosing between sentinel errors, custom types, and fmt.Errorf (%w vs %v), structuring error flow, or deciding whether to log or return. Also use when propagating errors across package boundaries or using errors.Is/As, even if the user doesn't ask about error strategy. Does not cover panic/recover patterns (see go-defensive).
allowed-tools: Bash(bash:*)

Go Error Handling

> Compatibility: `errors.Is`, `errors.As`, and `%w` wrapping require Go 1.13+; structured logging examples may use `log/slog` from Go 1.21+.

Resource Routing

• `scripts/check-errors.sh` - Run when checking string-based error matching, bare error propagation, and log-and-return patterns.
• `scripts/check-errors-ast.go` - Implementation helper invoked by `check-errors.sh`; patch this when changing error-flow analysis behavior.
• `references/ERROR-FLOW.md` - Read when deciding where to handle, wrap, log, or return errors.
• `references/ERROR-TYPES.md` - Read when choosing sentinel errors, typed errors, or opaque errors.
• `references/WRAPPING.md` - Read when choosing `%w` versus `%v` or crossing package boundaries.

In Go, [errors are values](https://go.dev/blog/errors-are-values) — they are created by code and consumed by code.

Choosing an Error Strategy

1. System boundary (RPC, IPC, storage)? → Wrap with `%v` to avoid leaking internals 2. Caller needs to match specific conditions? → Sentinel or typed error, wrap with `%w` 3. Caller just needs debugging context? → `fmt.Errorf("...: %w", err)` 4. Leaf function, no wrapping needed? → Return the error directly

**Default**: wrap with `%w` and place it at the end of the format string.

---

Core Rules

Never Return Concrete Error Types

**Never return concrete error types from exported functions** — a concrete `nil` pointer can become a non-nil interface:

// Bad: Concrete type can cause subtle bugs
func Bad() *os.PathError { /*...*/ }

// Good: Always return the error interface
func Good() error { /*...*/ }

Error Strings

Error strings should **not** be capitalized and should **not** end with punctuation. Exception: exported names, proper nouns, or acronyms.

// Bad
err := fmt.Errorf("Something bad happened.")

// Good
err := fmt.Errorf("something bad happened")

For displayed messages (logs, test failures, API responses), capitalization is appropriate.

Return Values on Error

When a function returns an error, callers must treat all non-error return values as unspecified unless explicitly documented.

**Tip**: Functions taking a `context.Context` should usually return an `error` so callers can determine if the context was cancelled.

---

Handling Errors

When encountering an error, make a **deliberate choice** — do not discard with `_`:

1. **Handle immediately** — address the error and continue 2. **Return to caller** — optionally wrapped with context 3. **In exceptional cases** — `log.Fatal` or `panic`

To intentionally ignore: add a comment explaining why.

n, _ := b.Write(p) // never returns a non-nil error

For related concurrent operations, use [`errgroup`](https://pkg.go.dev/golang.org/x/sync/errgroup):

g, ctx := errgroup.WithContext(ctx)
g.Go(func() error { return task1(ctx) })
g.Go(func() error { return task2(ctx) })
if err := g.Wait(); err != nil { return err }

Avoid In-Band Errors

Don't return `-1`, `nil`, or empty string to signal errors. Use multiple returns:

// Bad: In-band error value
func Lookup(key string) int  // returns -1 for missing

// Good: Explicit error or ok value
func Lookup(key string) (string, bool)

This prevents callers from writing `Parse(Lookup(key))` — it causes a compile-time error since `Lookup(key)` has 2 outputs.

---

Error Flow

Handle errors before normal code. Early returns keep the happy path unindented:

// Good: Error first, normal code unindented
if err != nil {
    return err
}
// normal code

**Handle errors once** — either log or return, never both:

Error encountered?
├─ Caller can act on it? → Return (with context via %w)
├─ Top of call chain? → Log and handle
└─ Neither? → Log at appropriate level, continue

---

Error Types

> **Advisory**: Recommended best practice.

| Caller needs to match? | Message type | Use | |------------------------|--------------|-----| | No | static | `errors.New("message")` | | No | dynamic | `fmt.Errorf("msg: %v", val)` | | Yes | static | `var ErrFoo = errors.New("...")` | | Yes | dynamic | custom `error` type |

**Default**: Wrap with `fmt.Errorf("...: %w", err)`. Escalate to sentinels for `errors.Is()`, to custom types for `errors.As()`.

---

Error Wrapping

> **Advisory**: Recommended best practice.

• **Use `%v`**: At system boundaries, for logging, to hide internal details
• **Use `%w`**: To preserve error chain for `errors.Is`/`errors.As`

**Key rules**: Place `%w` at the end. Add context callers don't have. If annotation adds nothing, return `err` directly.

> **Validation**: After implementing error handling, run `bash scripts/check-errors.sh` to detect common anti-patterns. Then run `go vet ./...` to catch additional issues.

---

Related Skills

• **Error naming**: See [go-naming](../go-naming/SKILL.md) when naming sentinel errors (`ErrFoo`) or custom error types
• **Testing errors**: See [go-testing](../go-testing/SKILL.md) when testing error semantics with `errors.Is`/`errors.As` or writing error-checking helpers
• **Panic handling**: See [go-defensive](../go-defensive/SKILL.md) when deciding between panic and error returns, or writing recover guards
• **Guard clauses**: See [go-control-flow](../go-control-flow/SKILL.md) when structuring early-return error flow or reducing nesting
• **Logging decisions**: See [go-logging](../go-logging/SKILL.md) when choosing log levels, configuring structured logging, or deciding what context to include in log messages